Translation result.
As generative artificial intelligence (AI) produces videos that are increasingly hard to distinguish from real footage, the content industry faces a central challenge: telling apart material shot by humans from what machines create. The world’s largest video platform has shifted from relying on creator self-reporting to using automated, platform-driven detection.
YouTube, owned by Google, said on its official blog on the 27th (local time) that it will roll out a feature to automatically identify and label AI-generated content. Until now, YouTube had asked creators to disclose realistic AI videos themselves; under the new policy, the system will add a label automatically when it detects substantial use of realistic AI, even if creators do not disclose it.
YouTube said it will rely on “internal signals” for detection — a measure meant to prevent creators from reverse-engineering the system and evading labels. The company described the change as an attempt to balance transparency with creators’ control.
Creators who believe a label was applied in error can change a video’s public status in YouTube Studio. However, videos produced with Google’s own AI tools — Veo or Dream Screen — or videos that include C2PA metadata, the open standard for provenance and authenticity, will retain AI-generation information permanently and cannot be unmarked.
The label’s placement will also become more prominent. For long-form videos, YouTube will display it directly below the player and above the description; for Shorts, the label will overlay the video. This visible labeling applies where AI has meaningfully or realistically altered or generated content. Unrealistic, animated, or lightly edited material will continue to include disclosures in the description. The 2024 requirement for creators to self-disclose remains in place, and AI labels will not affect a video’s recommendation ranking or its eligibility for monetization.
The move is part of a broader effort to curb low-quality AI content — the so-called “AI slop” problem. Last December, YouTube shut down two channels, Screen Culture and KH Studio, after they amassed more than a billion views with fake movie trailers — a step taken after Disney sent a cease-and-desist notice. On the 16th of last month, YouTube expanded its “Likeness Detection” feature, which finds videos that use AI to imitate a person’s face, to all creators age 18 and older.
Labeling efforts extend beyond YouTube. Spotify gives human creators an authenticity badge to distinguish them from AI-generated music, and Netflix has issued generative-AI guidelines to its production partners. Regulators are moving quickly, too. South Korea enacted the Framework Act on AI on Jan. 22, requiring disclosure when outputs are AI-generated and imposing fines of up to 30 million KRW (approximately $22,500) for violations. The Korea Fair Trade Commission also proposed revised review guidelines to require ads featuring AI-generated people to include wording such as “virtual person.”
The change illustrates that the default assumption that videos are “basically real” is eroding. If AI labels become routine, an unintended consequence may be that an absence of a label comes to signify authenticity. As virtual characters and AI-driven video proliferate across entertainment, proving a work’s origin and provenance has become a new competitive front.
Most Commented